Re: [boost] Regressions in your Boost libraries as of 2005-06-30

From: Paul Baxter (pauljbaxter_at_[hidden])
Date: 2005-07-02 14:58:24

> It looks like the "pjb" tests might be using an older revision of Intel
> 8.1 that's causing the trouble... I CC'd Paul in reply to Peter's
> message about bind (which is running into a similar problem). We'll try
> to get it cleaned up, but if we can't I can rein in the auto-complainer
> until we get it handled. Spurious errors are worse than no errors :(

Certainly its older in that it is a shameful :)  3 weeks out of date. 
v8.1.028
rather than latest which has recently had another bump.
I will rectify when I have a moment.

Could the problem be I'm running a release build rather than a debug one as
noted in my comments.html file

Anyone else run it in release to confirm/deny.

I won't be able to update now till Wednesday as my Intel login password is
at work and I'm not back till then. (I just tried and failed to remember my
password) Sorry for any delay in my help.

In the meantime I'll run a debug build with v8.1.028 tonight to see if I
duplicate other people's results and give back the (false?) sense of
security.

I did notice that I had PYTHON_VERSION set to 2.3 when in fact I'm running 
v2.4.1
I've modified that before running again.

Personally I think release build has more meaning for regressions but it
does have more chance of exposing compiler optimisation bugs rather than
boost ones....

Paul